added analysis, experiment, figures, updated readme

.gitignore

@@ -158,3 +158,4 @@ cython_debug/
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
 #.idea/
+.Rproj.user

README.md

@@ -4,7 +4,7 @@ Materials for the paper "[Whodunnit? Inferring what happened from multimodal evi
 
 Sarah A. Wu*, Erik Brockbank*, Hannah Cha, Jan-Philipp Fränken, Emily Jin, Zhuoyi Huang, Weiyu Liu, Ruohan Zhang, Jiajun Wu, Tobias Gerstenberg.
 
-Presented at the *46th Annual Conference of the Cognitive Science Society* (2024; Rotterdam, Netherlands).
+To be presented at the *46th Annual Conference of the Cognitive Science Society* (2024; Rotterdam, Netherlands).
 
 ```
 @inproceedings{wu2024whodunnit,
@@ -19,10 +19,9 @@ Presented at the *46th Annual Conference of the Cognitive Science Society* (2024
 **Contents:**
 
 * [Overview](#overview)
-* [Experiment pre-registration & demo](#experiment-preregistration--demo)
+* [Experiment](#experiment)
 * [Repository structure](#repository-structure)
-* [Experiment code](#experiment-code)
-* [GPT evaluation](#gpt-evaluation)
+* [GPT-4 evaluations](#gpt4-evaluations)
 * [Simulation model](#simulation-model)
 * [Generating trials](#generating-trials)
 * [CRediT author statement](#credit-author-statement)
@@ -45,45 +44,42 @@ Multimodal event reconstruction represents a challenge for current AI systems, a
 
 
 
-## Experiment preregistration & demo
+## Experiment
 
-The experiment reported in these results was pre-registered on the [Open Science Framework](https://help.osf.io/article/158-create-a-preregistration).
-
-Our pre-registration can be found [here](https://osf.io/fzxre).
-
-A demo of the experiment can be found [here](). <!--TODO add a demo link here -->
+The experiment reported in these results was pre-registered on the Open Science Framework [here](https://osf.io/fzxre).
+It can be previewed [here](https://cicl-stanford.github.io/whodunnit_multimodal_inference/experiment)!
 
 
 ## Repository structure
 
-
 ```
 ├── code
+│   ├── analysis
+│   ├── generate_audio
 │   ├── generate_visual
-│   │   └── ...
+│   ├── gpt4
 │   ├── model_data
-│   │   └── ...
 │   └── simulation_model
-│   │   └── ...
 ├── data
 ├── docs
+│   └── experiment
 ├── figures
 └── writeup
 ```
 
 - `/code`: This folder contains the code for various aspects of the experiment and analyses.
-    - `/model_data`: This folder has trial data in the format used by the simulation model and by GPT-4 and GPT-4V evaluations. The models use a combination of evidence images, scene graph JSON files, and a CSV with transcribed audio evidence for each trial.
+    - `/analysis`: contains all the code for analyzing data and generating figures (view a rendered file [here](https://cicl-stanford.github.io/whodunnit_multimodal_inference)).
+    - `/generate_visual`: contains code to generate the images for each trial from JSON specifications.
+    - `/generate_audio`: contains code to generate the audio files for each trial.
+    - `/gpt4`: This folder contains code to run GPT-4 and GPT-4V evaluations.
+    - `/model_data`: This folder has trial data in the format used by the simulation model and for GPT-4 and GPT-4V evaluations. The models use a combination of evidence images, scene graph JSON files, and a CSV with transcribed audio evidence for each trial.
     - `/simulation_model`: This folder has code and output for the [simulation model](#simulation-model).
-- `/data`:
-- `/docs`:
-- `/figures`:
-- `/writeup`:
+- `/data`: contains anonymized participant data from the experiment as well as GPT-4 and GPT-4V evaluation results.
+- `/docs/experiment`: contains all the behavioral experiment code. You can demo the experiment [here](https://cicl-stanford.github.io/whodunnit_multimodal_inference/experiment))!
+- `/figures`: contains all the figures from the paper, generated using the script in `code/analysis`.
 
 
-## Experiment code
-
-
-## GPT evaluation
+## GPT-4 evaluations
 
 
 ## Simulation model
@@ -199,7 +195,6 @@ This CSV file is used to compare participants' responses on each trial to the mo
 ## Generating trials
 
 
-
 ## CRediT author statement
 
 *[What is a CRediT author statement?](https://www.elsevier.com/researcher/author/policies-and-guidelines/credit-author-statement)*